基于Mahout协同过滤实现图书推荐系统-书籍推荐系统源码+项目说明.zip

package book.task; import book.dao.BookDao; import book.dao.BorrowDao; import book.dao.RecommendDao; import book.dao.UserDao; import book.domain.dataobject.*; import book.domain.exception.BusinessException; import book.util.LoggerUtil; import com.google.common.base.Joiner; import com.google.common.base.Splitter; import com.google.common.collect.Lists; import com.google.common.collect.Maps; import com.google.common.collect.Sets; import org.apache.commons.lang.StringUtils; import org.slf4j.Logger; import org.slf4j.LoggerFactory; import org.springframework.stereotype.Component; import org.springframework.transaction.support.TransactionTemplate; import org.springframework.util.CollectionUtils; import javax.annotation.Resource; import java.util.*; import static com.google.common.collect.Lists.newArrayList; /** * @author xlong * @date 2022-3-25 */ @Component public class KMeansRecommendation { private static final Logger LOGGER = LoggerFactory.getLogger(KMeansRecommendation.class); @Resource(name = "bookDao") private BookDao bookDao; @Resource(name = "userDao") private UserDao userDao; @Resource(name = "borrowDao") private BorrowDao borrowDao; @Resource(name = "recommendDao") private RecommendDao recommendDao; @Resource(name = "transactionTemplate") private TransactionTemplate transactionTemplate; private List<String> professionList; /* 1、用户注册时要求填写年龄，性别，职业； 2、首先计算各个职业之间的相似度P：将看过这本书且职业相同的人的评分求平均值，作为这个职业 （相当于一个用户，有多少个职业类别就有多少个用户）对这本书的评分，然后根据皮尔孙系数求相似度； 3、年龄之间的相似度A：7岁产生代沟的可能性较大，B(u,v)=7/(|Bu-Bv|),当|Bu-Bv|>7时；当|Bu-Bv|≤7时，B(u,v)=1; 4、性别相似度F：相同为0，不相同为1； 5、两个用户之间的距离即为A+F-P; */ public void KmeansRecommend() { LoggerUtil.info(LOGGER,"enter in KmeansRecommend"); recommendDao.deleteBefore(1); InitProfession(); //待分类的原始值 List<UserDO> userDOList=userDao.listAllUsers(); //将要分成的类别数 int k=(int)Math.sqrt(userDOList.size()); //定义最大迭代次数 int maxClusterTimes=500; //迭代次数 int count=1; //聚类的结果 List<List<UserDO>> clusterList=Lists.newArrayListWithCapacity(k); for (int i=0;i<k;i++){ clusterList.add(Lists.newArrayList()); } //质心 List<UserDO> clusterCenteringList=Lists.newArrayList(); //随机选择初始的质心 Collections.shuffle(userDOList); for(int i=0;i<k;i++) clusterCenteringList.add(userDOList.get(i)); //计算每个用户最靠近哪个质心，并将其加入相应的类中 boolean Change=true; while(Change&&count<=maxClusterTimes) { Change=false; for (int i = 0; i < userDOList.size(); i++) { //定义一个最小堆，存放用户到质心的距离 PriorityQueue<Map.Entry<Integer, Double>> priorityQueue = new PriorityQueue<Map.Entry<Integer, Double>>(new Comparator<Map.Entry<Integer, Double>>() { @Override public int compare(Map.Entry<Integer, Double> o1, Map.Entry<Integer, Double> o2) { return (int) (o1.getValue() - o2.getValue()); } }); Map<Integer, Double> map = Maps.newHashMap(); for (int j = 0; j < k&&j<clusterCenteringList.size(); j++) { double distance = CalDistance(userDOList.get(i), clusterCenteringList.get(j)); map.put(j, distance); } priorityQueue.addAll(map.entrySet()); //从堆顶获得最短距离对应的质心下标，将用户加到对应的质心的那一类中 clusterList.get(priorityQueue.poll().getKey()).add(userDOList.get(i)); } System.out.println(count); List<UserDO> newCenterList = CenterUpdate(clusterList); if(isChange(clusterCenteringList,newCenterList)) { Change=true; clusterCenteringList=newCenterList; } count++; System.out.println(count); } System.out.println(clusterList); //销量的最大堆 Queue<Map.Entry<String,Integer>> priorityQueue=new PriorityQueue<>((a,b)->(b.getValue().compareTo(a.getValue()))); Map<String, Integer> bookCountMap=CountSale(); priorityQueue.addAll(bookCountMap.entrySet()); //遍历每一个用户，找出他属于哪一类 //对于这一类中其他用户看过的书，求并集去重 //然后这些书和他看过的书求差集，存放在最小堆里面 for(UserDO userDO:userDOList) { //推荐书籍的数量 int recommendCount=0; priorityQueue.clear(); //兜底，每次都重新计算借阅量最大堆，防止为这个用户推荐的图书不够三本 priorityQueue.addAll(bookCountMap.entrySet()); //记录用户所在类的下标 int ClusterIndex=0; //扫描每一个类 for(int i=0;i<clusterList.size();i++) { //扫描这一类中的用户 for(int j=0;j<clusterList.get(i).size();j++) { if(userDO.getUserId()==clusterList.get(i).get(j).getUserId()) { ClusterIndex=i; break; } } } //存放其他其他用户看过的书：bookName-author集合 Set<String> bookSet=Sets.newHashSet(); //该用户自己已经看过的书 Set<String> MyBookSet=Sets.newHashSet(); for(UserDO userDO1:clusterList.get(ClusterIndex)) { //计算和该用户同一类的其他用户看过的书籍 if(userDO.getUserId()!=userDO1.getUserId()) { List<BorrowDO> borrowDOList=borrowDao.listByUserId(userDO1.getUserId()); for(BorrowDO borrowDO:borrowDOList) { BookDO bookDO=bookDao.queryBookByBookId(borrowDO.getBookId()); bookSet.add(Joiner.on("-").skipNulls().join(bookDO.getBookName(),bookDO.getAuthor())); } } //计算本人已经看过的书籍 else { List<BorrowDO> borrowDOList=borrowDao.listByUserId(userDO.getUserId()); for(BorrowDO borrowDO:borrowDOList) { BookDO bookDO=bookDao.queryBookByBookId(borrowDO.getBookId()); MyBookSet.add(Joiner.on("-").skipNulls().join(bookDO.getBookName(),bookDO.getAuthor())); } } } //求出bookSet中有的而MyBookSet中没有的书 bookSet=Sets.difference(bookSet,MyBookSet); List<BookDO> bookDOList=Lists.newArrayList(); for(String bookAuthor:bookSet) { //分离出书名和作者 Iterator<String> iterator = Splitter.on("-").split(bookAuthor).iterator(); List<String> stringList = Lists.newArrayList(); while (iterator.hasNext()) { stringList.add(iterator.next()); } //根据书名和作者名查询这本书 List<BookDO> bookDOList1=bookDao.listBooksByAuthorAndName(stringList.get(0),stringList.get(1));